Objective: The aim of this study was to evaluate the actual dose variability to the targets and organs at risk (OARs) during nasopharyngeal carcinoma (NPC) intensitymodulated radiotherapy (IMRT) and to investigate the significance of replanning. Methods: 11 NPC patients were included in this study. Each patient had both a planning CT and weekly repeated CT. Simulated plans that were generated by using the same beam configurations mapped to the repeated CT represented the actual delivered doses to the target volumes and OARs. An IMRT replanning was performed with the fifth week CT scan. Doses among the initial plan, the simulated plans and replanning were compared. Results: There were no significant dosimetric differences in the gross tumour volume, clinical target volume (CTV) 1 or CTV2 for either the simulated plans or the replanning compared with the initial plan. Dosimetric variability of both parotid glands and the brain stem were unique to each individual, and doses to the spinal cord were always maintained within the limit. Replanning in the fifth week had significantly decreased the doses delivered to both parotids (p-values of the mean dose were 0.015 and 0.026 for the left and right parotid, respectively), whereas it did not reduce the doses to the brain stem and spinal cord. There was no relationship between dose variability and weight loss. Conclusions: There are no significant dose changes for target volumes and spinal cord, and doses to the brain stem and both parotid glands changed individually during NPC IMRT. Replanning helps to spare bilateral parotids.